BK Housing Info

BK Housing Info is where the data in ComfyUI_Baikong_Buying begins - a form masquerading as a node. You fill in the details of a target property (in Chinese, like everything in this pack), and out comes a single CSV row, ready to be rendered, sorted, or chained into a shortlist. It does no math you couldn't do in a spreadsheet; what it does is turn a listing's details into the pack's CSV format so the rest of the workflow can handle it.

What it is and why you'd bother

This pack, by JayLyu ("baikong"), uses ComfyUI as a home-buying decision tool - no models, no GPU, no API keys, just pandas, matplotlib, and a lot of widgets. BK_HousingInfo is the input side. Its one output, CSV, feeds the pack's terminal node, BK Housing Calc, which sorts the table and renders it as an image - or you can chain several BK_HousingInfo nodes together to accumulate a multi-property shortlist. The standalone BK Housing Decision node is where the actual buy/don't-buy numbers happen.

The inputs

Every input is optional and has a default, so the node works with zero wiring. The form covers the fields a Chinese property listing cares about:

• The basics: 小区名称 (community), 房屋总价 (price, in yuan), 室/厅/卫 (rooms/halls/baths), 建筑面积 (floor area in m²), 得房率 (usable-area ratio, as a percentage).
• The building: 当前楼层/总楼层, 装修标准 (renovation standard), 容积率 (plot ratio), 总户数 (households).
• The enums - worth knowing because they're dropdowns, not free text: 房屋权属 (商品房/公房/经济适用房/其他), 楼房类型 (塔楼/板楼/高层/小高层/超高层), 房屋用途 (普通住宅/商业类/别墅/四合院/车位/其他), 房屋类型 (新房/二手房), 装修 (精装修/无).
• 交房时间 and 物业费 (monthly property fee per m²).

How it works

Under the hood it builds a one-row pandas DataFrame and formats the fields into human-friendly strings: price becomes 420.00万元, the floor plan becomes 3室2厅2卫, and it derives 套内面积 (usable area = 建筑面积 × 得房率/100) and yearly property fee (物业费 × 面积 × 12). The genuinely useful bit is the optional pipeline input, typed CSV: wire a previous BK_HousingInfo's CSV into it and the new row is appended to that table instead of starting fresh. Two or three nodes in a chain, and you've built a multi-property comparison list without touching a line of code. (The CSV socket type only mates with other nodes in this pack - that's normal; it's the pack's private data bus.)

Installing it

One pack, one install: ComfyUI Manager → search "ComfyUI_Baikong_Buying" → install → restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/JayLyu/ComfyUI_Baikong_Buying

requirements.txt pulls torch, pandas, numpy, matplotlib, seaborn; torch/numpy ship with ComfyUI and the rest are standard. No model downloads, no font to find (the CJK font is bundled). The whole pack is CPU-only.

Gotchas

The defaults are a real demo listing - 滨运锦绣里, ¥4.2M, 3室2厅2卫 - so the first thing to do is change every field you care about, or you'll be compiling a shortlist of someone else's imaginary apartment. The node's output text is entirely Chinese, and the only terminal feedback is a debug line in the console (e.g. [BK_District_Info] ○ INPUT 滨运锦绣里-3室2厅2卫). If you're not reading Chinese, keep a translator open for the field names - the values themselves are straightforward numbers once you've mapped them once. And remember this node is just the form: wire its CSV into BK_HousingCalc to actually see the table.